Power yaw control for distributed electric propulsion aircraft based on reinforcement learning

Distributed electric propulsion aircraft provides thrust through multiple electric thrusters distributed on the wing. This design utilises the thrust differential generated by the distributed electric thrusters to achieve power yaw， thereby improving flight efficiency. To achieve power yaw control of the aircraft， it is necessary to ensure that the distributed electric thrusters operate cooperatively with each other. Therefore， this paper proposed a power yaw control method for distributed electric propulsion aircraft based on Deep Deterministic Policy Gradient （DDPG） reinforcement learning algorithm. And based on NASA X-57 aircraft， a nonlinear equivalent scaled aircraft model is established to make the training environment of the algorithm closer to the real flight environment. The results show that the method is able to achieve reasonable distribution of thrust among distributed electric thrusters， thus realising power yaw control. Compared with the traditional control method， the proposed method reduces the yaw response settling time of the aircraft by 36.36% while ensuring flight stability.